123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413 |
- <?php
- class VkBridge extends BridgeAbstract
- {
- const MAINTAINER = 'ahiles3005';
- const NAME = 'VK.com';
- const URI = 'https://vk.com/';
- const CACHE_TIMEOUT = 300; // 5min
- const DESCRIPTION = 'Working with open pages';
- const PARAMETERS = array(
- array(
- 'u' => array(
- 'name' => 'Group or user name',
- 'required' => true
- )
- )
- );
- protected $videos = array();
- protected $pageName;
- protected function getAccessToken()
- {
- return 'c8071613517c155c6cfbd2a059b2718e9c37b89094c4766834969dda75f657a2c1cbb49bab4c5e649f1db';
- }
- public function getURI()
- {
- if (!is_null($this->getInput('u'))) {
- return static::URI . urlencode($this->getInput('u'));
- }
- return parent::getURI();
- }
- public function getName()
- {
- if ($this->pageName) {
- return $this->pageName;
- }
- return parent::getName();
- }
- public function collectData()
- {
- $text_html = $this->getContents()
- or returnServerError('No results for group or user name "' . $this->getInput('u') . '".');
- $text_html = iconv('windows-1251', 'utf-8', $text_html);
- // makes album link generating work correctly
- $text_html = str_replace('"class="page_album_link">', '" class="page_album_link">', $text_html);
- $html = str_get_html($text_html);
- $pageName = $html->find('.page_name', 0);
- if (is_object($pageName)) {
- $pageName = $pageName->plaintext;
- $this->pageName = htmlspecialchars_decode($pageName);
- }
- foreach ($html->find('div.replies') as $comment_block) {
- $comment_block->outertext = '';
- }
- $html->load($html->save());
- $pinned_post_item = null;
- $last_post_id = 0;
- foreach ($html->find('.post') as $post) {
- defaultLinkTo($post, self::URI);
- $post_videos = array();
- $is_pinned_post = false;
- if (strpos($post->getAttribute('class'), 'post_fixed') !== false) {
- $is_pinned_post = true;
- }
- if (is_object($post->find('a.wall_post_more', 0))) {
- //delete link "show full" in content
- $post->find('a.wall_post_more', 0)->outertext = '';
- }
- $content_suffix = '';
- // looking for external links
- $external_link_selectors = array(
- 'a.page_media_link_title',
- 'div.page_media_link_title > a',
- 'div.media_desc > a.lnk',
- );
- foreach($external_link_selectors as $sel) {
- if (is_object($post->find($sel, 0))) {
- $a = $post->find($sel, 0);
- $innertext = $a->innertext;
- $parsed_url = parse_url($a->getAttribute('href'));
- if (strpos($parsed_url['path'], '/away.php') !== 0) continue;
- parse_str($parsed_url['query'], $parsed_query);
- $content_suffix .= "<br>External link: <a href='" . $parsed_query['to'] . "'>$innertext</a>";
- }
- }
- // remove external link from content
- $external_link_selectors_to_remove = array(
- 'div.page_media_thumbed_link',
- 'div.page_media_link_desc_wrap',
- 'div.media_desc > a.lnk',
- );
- foreach($external_link_selectors_to_remove as $sel) {
- if (is_object($post->find($sel, 0))) {
- $post->find($sel, 0)->outertext = '';
- }
- }
- // looking for article
- $article = $post->find('a.article_snippet', 0);
- if (is_object($article)) {
- if (strpos($article->getAttribute('class'), 'article_snippet_mini') !== false) {
- $article_title_selector = 'div.article_snippet_mini_title';
- $article_author_selector = 'div.article_snippet_mini_info > .mem_link,
- div.article_snippet_mini_info > .group_link';
- $article_thumb_selector = 'div.article_snippet_mini_thumb';
- } else {
- $article_title_selector = 'div.article_snippet__title';
- $article_author_selector = 'div.article_snippet__author';
- $article_thumb_selector = 'div.article_snippet__image';
- }
- $article_title = $article->find($article_title_selector, 0)->innertext;
- $article_author = $article->find($article_author_selector, 0)->innertext;
- $article_link = $article->getAttribute('href');
- $article_img_element_style = $article->find($article_thumb_selector, 0)->getAttribute('style');
- preg_match('/background-image: url\((.*)\)/', $article_img_element_style, $matches);
- if (count($matches) > 0) {
- $content_suffix .= "<br><img src='" . $matches[1] . "'>";
- }
- $content_suffix .= "<br>Article: <a href='$article_link'>$article_title ($article_author)</a>";
- $article->outertext = '';
- }
- // get video on post
- $video = $post->find('div.post_video_desc', 0);
- $main_video_link = '';
- if (is_object($video)) {
- $video_title = $video->find('div.post_video_title', 0)->plaintext;
- $video_link = $video->find('a.lnk', 0)->getAttribute('href');
- $this->appendVideo($video_title, $video_link, $content_suffix, $post_videos);
- $video->outertext = '';
- $main_video_link = $video_link;
- }
- // get all other videos
- foreach($post->find('a.page_post_thumb_video') as $a) {
- $video_title = htmlspecialchars_decode($a->getAttribute('aria-label'));
- $temp = explode(' ', $video_title, 2);
- if (count($temp) > 1) $video_title = $temp[1];
- $video_link = $a->getAttribute('href');
- if ($video_link != $main_video_link) $this->appendVideo($video_title, $video_link, $content_suffix, $post_videos);
- $a->outertext = '';
- }
- // get all photos
- foreach($post->find('div.wall_text > a.page_post_thumb_wrap') as $a) {
- $result = $this->getPhoto($a);
- if ($result == null) continue;
- $a->outertext = '';
- $content_suffix .= "<br>$result";
- }
- // get albums
- foreach($post->find('.page_album_wrap') as $el) {
- $a = $el->find('.page_album_link', 0);
- $album_title = $a->find('.page_album_title_text', 0)->getAttribute('title');
- $album_link = $a->getAttribute('href');
- $el->outertext = '';
- $content_suffix .= "<br>Album: <a href='$album_link'>$album_title</a>";
- }
- // get photo documents
- foreach($post->find('a.page_doc_photo_href') as $a) {
- $doc_link = $a->getAttribute('href');
- $doc_gif_label_element = $a->find('.page_gif_label', 0);
- $doc_title_element = $a->find('.doc_label', 0);
- if (is_object($doc_gif_label_element)) {
- $gif_preview_img = backgroundToImg($a->find('.page_doc_photo', 0));
- $content_suffix .= "<br>Gif: <a href='$doc_link'>$gif_preview_img</a>";
- } else if (is_object($doc_title_element)) {
- $doc_title = $doc_title_element->innertext;
- $content_suffix .= "<br>Doc: <a href='$doc_link'>$doc_title</a>";
- } else {
- continue;
- }
- $a->outertext = '';
- }
- // get other documents
- foreach($post->find('div.page_doc_row') as $div) {
- $doc_title_element = $div->find('a.page_doc_title', 0);
- if (is_object($doc_title_element)) {
- $doc_title = $doc_title_element->innertext;
- $doc_link = $doc_title_element->getAttribute('href');
- $content_suffix .= "<br>Doc: <a href='$doc_link'>$doc_title</a>";
- } else {
- continue;
- }
- $div->outertext = '';
- }
- // get polls
- foreach($post->find('div.page_media_poll_wrap') as $div) {
- $poll_title = $div->find('.page_media_poll_title', 0)->innertext;
- $content_suffix .= "<br>Poll: $poll_title";
- foreach($div->find('div.page_poll_text') as $poll_stat_title) {
- $content_suffix .= '<br>- ' . $poll_stat_title->innertext;
- }
- $div->outertext = '';
- }
- // get sign
- $post_author = $pageName;
- foreach($post->find('a.wall_signed_by') as $a) {
- $post_author = $a->innertext;
- $a->outertext = '';
- }
- if (is_object($post->find('div.copy_quote', 0))) {
- $copy_quote = $post->find('div.copy_quote', 0);
- if ($copy_post_header = $copy_quote->find('div.copy_post_header', 0)) {
- $copy_post_header->outertext = '';
- }
- $copy_quote_content = $copy_quote->innertext;
- $copy_quote->outertext = "<br>Reposted: <br>$copy_quote_content";
- }
- $item = array();
- $item['content'] = strip_tags(backgroundToImg($post->find('div.wall_text', 0)->innertext), '<br><img>');
- $item['content'] .= $content_suffix;
- $item['categories'] = array();
- // get post hashtags
- foreach($post->find('a') as $a) {
- $href = $a->getAttribute('href');
- $prefix = '/feed?section=search&q=%23';
- $innertext = $a->innertext;
- if ($href && substr($href, 0, strlen($prefix)) === $prefix) {
- $item['categories'][] = urldecode(substr($href, strlen($prefix)));
- } else if (substr($innertext, 0, 1) == '#') {
- $item['categories'][] = $innertext;
- }
- }
- // get post link
- $post_link = $post->find('a.post_link', 0)->getAttribute('href');
- preg_match('/wall-?\d+_(\d+)/', $post_link, $preg_match_result);
- $item['post_id'] = intval($preg_match_result[1]);
- $item['uri'] = $post_link;
- $item['timestamp'] = $this->getTime($post);
- $item['title'] = $this->getTitle($item['content']);
- $item['author'] = $post_author;
- $item['videos'] = $post_videos;
- if ($is_pinned_post) {
- // do not append it now
- $pinned_post_item = $item;
- } else {
- $last_post_id = $item['post_id'];
- $this->items[] = $item;
- }
- }
- if (!is_null($pinned_post_item)) {
- if (count($this->items) == 0) {
- $this->items[] = $pinned_post_item;
- } else if ($last_post_id < $pinned_post_item['post_id']) {
- $this->items[] = $pinned_post_item;
- usort($this->items, function ($item1, $item2) {
- return $item2['post_id'] - $item1['post_id'];
- });
- }
- }
- $this->getCleanVideoLinks();
- }
- private function getPhoto($a) {
- $onclick = $a->getAttribute('onclick');
- preg_match('/return showPhoto\(.+?({.*})/', $onclick, $preg_match_result);
- if (count($preg_match_result) == 0) return;
- $arg = htmlspecialchars_decode( str_replace('queue:1', '"queue":1', $preg_match_result[1]) );
- $data = json_decode($arg, true);
- if ($data == null) return;
- $thumb = $data['temp']['base'] . $data['temp']['x_'][0] . '.jpg';
- $original = '';
- foreach(array('y_', 'z_', 'w_') as $key) {
- if (!isset($data['temp'][$key])) continue;
- if (!isset($data['temp'][$key][0])) continue;
- if (substr($data['temp'][$key][0], 0, 4) == 'http') {
- $base = '';
- } else {
- $base = $data['temp']['base'];
- }
- $original = $base . $data['temp'][$key][0] . '.jpg';
- }
- if ($original) {
- return "<a href='$original'><img src='$thumb'></a>";
- } else {
- return "<img src='$thumb'>";
- }
- }
- private function getTitle($content)
- {
- preg_match('/^["\w\ \p{Cyrillic}\(\)\?#«»-]+/mu', htmlspecialchars_decode($content), $result);
- if (count($result) == 0) return 'untitled';
- return $result[0];
- }
- private function getTime($post)
- {
- if ($time = $post->find('span.rel_date', 0)->getAttribute('time')) {
- return $time;
- } else {
- $strdate = $post->find('span.rel_date', 0)->plaintext;
- $date = date_parse($strdate);
- if (!$date['year']) {
- if (strstr($strdate, 'today') !== false) {
- $strdate = date('d-m-Y') . ' ' . $strdate;
- } elseif (strstr($strdate, 'yesterday ') !== false) {
- $time = time() - 60 * 60 * 24;
- $strdate = date('d-m-Y', $time) . ' ' . $strdate;
- } else {
- $strdate = $strdate . ' ' . date('Y');
- }
- $date = date_parse($strdate);
- }
- return strtotime($date['day'] . '-' . $date['month'] . '-' . $date['year'] . ' ' .
- $date['hour'] . ':' . $date['minute']);
- }
- }
- private function getContents()
- {
- ini_set('user-agent', 'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:53.0) Gecko/20100101 Firefox/53.0');
- $header = array('Accept-language: en', 'Cookie: remixlang=3');
- return getContents($this->getURI(), $header);
- }
- protected function appendVideo($video_title, $video_link, &$content_suffix, array &$post_videos)
- {
- if (!$video_title) $video_title = '(empty)';
- preg_match('/video([0-9-]+_[0-9]+)/', $video_link, $preg_match_result);
- if (count($preg_match_result) > 1) {
- $video_id = $preg_match_result[1];
- $this->videos[ $video_id ] = array(
- 'url' => $video_link,
- 'title' => $video_title,
- );
- $post_videos[] = $video_id;
- } else {
- $content_suffix .= '<br>Video: <a href="'.htmlspecialchars($video_link).'">'.$video_title.'</a>';
- }
- }
- protected function getCleanVideoLinks() {
- $result = $this->api('video.get', array(
- 'videos' => implode(',', array_keys($this->videos)),
- 'count' => 200
- ));
- if (isset($result['error'])) return;
- foreach($result['response']['items'] as $item) {
- $video_id = strval($item['owner_id']).'_'.strval($item['id']);
- $this->videos[$video_id]['url'] = $item['player'];
- }
- foreach($this->items as &$item) {
- foreach($item['videos'] as $video_id) {
- $video_link = $this->videos[$video_id]['url'];
- $video_title = $this->videos[$video_id]['title'];
- $item['content'] .= '<br>Video: <a href="'.htmlspecialchars($video_link).'">'.$video_title.'</a>';
- }
- unset($item['videos']);
- }
- }
- protected function api($method, array $params)
- {
- $params['v'] = '5.80';
- $params['access_token'] = $this->getAccessToken();
- return json_decode( getContents('https://api.vk.com/method/'.$method.'?'.http_build_query($params)), true );
- }
- }
|